• wireless sensor networks
• ADL
• accelerometer
• change detection
• rotation matrices
• Activities of Daily Living

During the last century, mortality decreased significantly. Health-related behavior, physi-
cal activity, nutrition and socio-economic conditions allow people to have a great life ex-
pectancy.
As consequence of this situation, we have an increase of the average population age. In
fact, today it is not rare to meet an over-centenary person, and in the following years it
will be even normal.
Nevertheless elderly people needs assistance and in some cases it must be continuous
(for example, for the people affected by Alzheimer's Disease). In this context, monitoring
the Activities of Daily Living (ADLs) of elderly people becomes more and more impor -
tant. Systems that make possible this monitoring have the aim to keep track of activities
that are executed during a normal life day, in order to known patient's health state and
identify dangerous situations.
Among the ADLs Monitoring Systems, those dedicated to Posture Detection play an im-
portant role. These systems have as a target the identification of the patient's posture
and they help to obtain information useful for an analysis of his health in collaboration
with other detection systems.
The development of an autonomous and low invasive Posture Detection System based
on Wireless Sensor Networks is the aim of this thesis. One sensor (also called 'mote') is
worn by the patient on his lower abdomen. It is equipped with a three-axis accelerome-
ter.By analyzing the acceleration trend, it is possible to observe that different postures have
different behaviors in terms of acceleration. These behaviors suggest us to use a new
approach, which is based on the use of Change Detection algorithms, that can identify
particular behaviors with a low computational overhead. Change Detection Algorithms
have been used from long time in various fields, such as quality control processes,
economy, planes control systems and many other sectors.